No . because that criteria is restricted by your software to models that have s set = p-1
In general s is NOT tied to p thus any simplified implicit specification that s === p-1 is generally false and is used because the pre-set set of models is limited. and because there may be Level Shifts , Local Time Trends , Pulses and/or Seasonal Pulses in your data which need to be accommodate ( accounted for ) by deterministic input series . http://www.autobox.com/pdfs/catchword.pdf Hope this helps . Dave Reilly Automatic Forecasting Systems http://www.autobox.com -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of C. Cheung Sent: Thursday, April 01, 2004 6:36 AM To: [EMAIL PROTECTED] Subject: [edstat] Lowest AIC=the best ARIMA/SARIMA model? "the model with the lowest AIC is the best model" is it always true for ARIMA or SARIMA model? . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . ================================================================= . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
